Van's response to said post. "Sander,
I'm not going to get into a flame war with you, but if you persist I will publish every email I have from you and let people decide who is on the level (or even who is sane). I am weary of arguing.
For everyone else, Sander sent me a number of unsolicited, harassing emails about my Rambus articles (and perhaps for one or two other articles) and if I remember correctly, Sander even tried to get me in trouble with Tom.
Sander also posted a series of counter articles on HardwareCentral that were painfully stilted. His own readership skewered (http://discussions.hardwarecentral.com/Forum15/HTML/000016.html -- note the irony of the article's title) him for these pieces -- articles written with clear agendas. AND, Sander was simply wrong as time has shown.
Why did Sander harass me and write these skewed counter arguments?
Sander, for some bizarre reason, you have seemed to have tried to take the mantel of my nemesis -- and you can't resist getting a kick to a kidney as I'm on my way out.
However, you have been largely unsuccessful in your attacks because I have tried to act out of honesty, I have always tried to be truthful.
Sander, don't measure yourself against me or anyone else. Just do what is right.
There is little doubt of your friendly relationships with Intel and Rambus -- as a matter of fact I remember your telling me at fall's IDF about your many contacts at Rambus and how you could set me up with one. You also seem to be close to Intel's George Alfs (here's your man George – I'm sure you're both making Intel proud – I’m also sure you both can stand tall before God when that time comes). You were very proud of your strong connections with these two companies.
I realize the market is lean right now, and it is turning even a few earnest people to compromise for ad revenue. But ad money wasn't tight when you were flaming me through email about my Rambus pieces. I do not understand your warped perspective.
If I were in your position as “CEO” of a website, I would certainly be taking a different tack than yours. You have been blessed with your opportunity – opportunity that has come twice for you already.
However, I would rather leave this business than to deceive and manipulate readers -- or even remotely be a part of any attempt to do so. Certainly I’m not perfect, but this has always been my conviction.
So you do have a choice Sander.
"
aceshardware.com
V.S. Writes "As I've posted here before in some detail, there are two primary throttling mechanisms at work with the P4. There is automatic hardware throttling that effectively forces a 50% StopClock duty cycle on the chip once a certain temperature (not clearly disclosed) is exceeded at the Thermal Monitor's temperature diode.
Please note this temperature is not reflected in Intel's monitoring program or any other program that I am aware of. Some programs do use the integrated legacy thermal diode (which still triggers a PIII style shutdown if the chip overheats), but this sensor does not give any insight into throttling.
In fact, the Thermal Monitor's sensor is embedded in an area of the P4 prone to hot spots. This means that the temperatures it measures almost certainly vary much more than the legacy diode's (which is used to obtain an idea of generalized chip temperature).
According to Intel documentation, P4 hot spot throttling can occur almost instantaneously. Although other chips are naturally prone to hot spots, the P4's issues seem to be more acute due to a combination of factors:
1. large die sizes are more prone to large temperature gradients.
2. Intel's reliance on "massive" clock gating in the P4 to reduce typical power consumption increases temperature gradients and lowers the overall die temperature reflected by legacy thermal diode.
3. Intel's double-pumped ALU (originally code named "FireBall," I believe) and other architectural design desisions can promote localized overheating.
Besides the automatic hardware throttling, the P4 has a much more subtle throttling mechanism that can be used with ACPI and other software mechanisms. This mode of throttling is controlled by three bits which degrade performance in steps of 12.5%.
We believe the reason that we saw one 1.7GHz P4 throttle on QIII on both Intel D850GB motherboards, but not on an ABIT motherboard is due to a more conservative BIOS in the Intel motherboard.
We witnessed QIII throttling with one 1.7GHz P4, but not with another. Both were retail versions of the chip.
Here is a link to a posting I made last month on this issue. You will also find in that posting a link to the Intel document.
aceshardware.com
---------------------------------------------------------------
aceshardware.com |
